Generative Refinement Networks for Visual Synthesis
This work addresses the computational inefficiency of diffusion models and the quality bottlenecks of autoregressive models in visual generation, offering a more efficient and high-quality alternative for practitioners.
Generative Refinement Networks (GRN) introduce a new visual synthesis paradigm that combines near-lossless Hierarchical Binary Quantization with global refinement and entropy-guided sampling, achieving state-of-the-art image reconstruction (0.56 rFID) and class-conditional generation (1.81 gFID) on ImageNet, and superior performance in text-to-image and text-to-video tasks.
While diffusion models dominate the field of visual generation, they are computationally inefficient, applying a uniform computational effort regardless of different complexity. In contrast, autoregressive (AR) models are inherently complexity-aware, as evidenced by their variable likelihoods, but are often hindered by lossy discrete tokenization and error accumulation. In this work, we introduce Generative Refinement Networks (GRN), a next-generation visual synthesis paradigm to address these issues. At its core, GRN addresses the discrete tokenization bottleneck through a theoretically near-lossless Hierarchical Binary Quantization (HBQ), achieving a reconstruction quality comparable to continuous counterparts. Built upon HBQ's latent space, GRN fundamentally upgrades AR generation with a global refinement mechanism that progressively perfects and corrects artworks -- like a human artist painting. Besides, GRN integrates an entropy-guided sampling strategy, enabling complexity-aware, adaptive-step generation without compromising visual quality. On the ImageNet benchmark, GRN establishes new records in image reconstruction (0.56 rFID) and class-conditional image generation (1.81 gFID). We also scale GRN to more challenging text-to-image and text-to-video generation, delivering superior performance on an equivalent scale. We release all models and code to foster further research on GRN.